You are on page 1of 38

Insurance for the Poor?

Stefan Dercon
Tessa Bold
César Calvo

Inter-American Development Bank

Washington, D.C.

Sustainable Development Department
Technical Paper Series
Stefan Dercon is Professor of Economics at Oxford University, Tessa Bold is a Junior Research Fellow at
Oxford University, César Calvo is a Departmental Teaching Associate at Oxford University. The authors
would like to thank Luis Tejerina for his insightful comments.

The opinions expressed herein are those of the authors and do not necessarily represent the official posi-
tion of the Inter-American Development Bank. Permission is granted to reproduce this paper for non-
commercial purposes only and with proper attribution to the authors, the Sustainable Development De-
partment and the Bank.

Publication of the Inter-American Development Bank, August, 2006

Manager a.i, Sustainable Development Department: Antonio Vives
Chief, Poverty and Inequality Unit: Carlos Eduardo Vélez
Chief, Infrastructure and Financial Services: Pietro Masci

For additional copies of this publication (Reference No. IFM-145), please contact:

Poverty and Inequality Unit Infrastructure and Financial Markets Division
Sustainable Development Department Sustainable Development Department
Inter-American Development Bank Inter-American Development Bank
1300 New York Avenue, N.W. 1300 New York Avenue, N.W.
Washington, D.C. 20577 Washington, D.C. 20577

Email: or
Fax: 202-623-2157
Web site:

Cataloging -in-Publication provided by the
Inter-American Development Bank
Felipe Herrera Library

Dercon, Stefan.

Insurance for the poor? / Stefan Dercon, Tessa Bold, César Calvo. (Sustainable Development Department Technical papers series ; IFM-145)
Includes bibliographical references.

1. (Risk) Insurance—Latin America. 2. Poor—Latin America. I. Bold, Tessa. II. Calvo, César. III.
Inter-American Development Bank. Sustainable Development Dept. Infrastructure and Financial Mar-
kets Division. IV. Title. V. Series.

HG8054.5 D27 2006
368 D27—dc22

Following the Eighth Replenishment in 1994, the Inter-American Development Bank identified poverty
reduction and social equity promotion, and sustainable economic growth as its two overarching strategic
goals. Undoubtedly, more pro-poor and equitable access to financial markets promotes both strategic
goals. Accordingly, the Bank has been providing support to sector reforms and investment projects to
strengthen financial markets and to improve access to credit and know-how to small, medium and micro-
entrepreneurs throughout the region.

The Bank’s involvement in the development of financial markets was strengthened by the conference
“Challenges and Opportunities for the Development of Insurance Markets in Latin America and the Car-
ibbean,” which the Bank organized in November, 2005. The conference brought together people involved
in the region’s insurance sector (industry, regulators, and users) to discuss and identify the factors that
make insurance effective and agree upon a working agenda for the future. Moreover, the initiative “Build-
ing Opportunities for the Majority,” which was launched in 2006, highlights “financial democracy” as one
priority area for Bank activity in the near future.

Contrary to the industry’s traditional practices, financial institutions are now paying more attention to the
use of financial instruments as risk management tools for the poor. Available evidence suggests that in
order to manage risk and cope against idiosyncratic shocks, low-income households are using credit and
savings from various sources, and that a wide range of formal insurance mechanisms have been used in
many countries with various degrees of success. It is expected that these mechanisms will enable low-
income households to make investments with higher expected rates of return, and in times of need (or
negative economic shocks) will provide them with welfare protection options and avoid costly coping
mechanisms, such as asset depletion or borrowing at high interest rates.

This study was prepared for the Financial Services and Poverty Reduction research project and was pre-
sented at the conference with the same name held at Bank headquarters in September of 2004. It first es-
tablishes the considerable welfare cost borne by the region’s poor when facing uncovered risks (which, in
principle, are insurable), and offers empirical evidence that shows the clear disadvantages of the informal
risk management instruments typically used by the poor. Next, the paper proposes ways to increase access
to insurance for the poor in the region. Particularly relevant, among the proposed solutions, is the imple-
mentation of the partner-agent model in which formal insurance institutions are coupled with existing mi-
crofinance institutions for the delivery of their products, making the most of the comparative advantages
of each sector.

While the use of microinsurance should not be looked upon as a “silver bullet” solution to all types of
risks management problems, it is clear that access to this type of instrument can improve the welfare of
the poor by providing access to a wider pool of potential investments and is also a cost effective instru-
ments for coping with negative shocks.

Pietro Masci Carlos Eduardo Vélez
Chief, Infrastructure and Chief, Poverty and Inequality Unit
Financial Markets Division

Introduction 1

Risk and the Poor 3

Market Failures and Household Responses to Risk 8

Scope for Insurance Provision to the Poor 12

Implementing Insurance for the Poor 16

Insurance Products for the Poor 18

Regulation for Insurance Provision to the Poor 23

Local Institutions and Insurance Provision to the Poor 25

Conclusions 28

References 30

Households in developing countries are exposed complementarities with other mechanisms to
to high risks, with important consequences on reduce and to cope with risk, including for ex-
their welfare. They range from individual- ample, safety nets in the form of employment
specific (illness, theft or unemployment), to schemes or social funds. A related issue, and
economy-wide risks (drought, recession, etc.). It crucial for Latin America, is that much of the
has long been acknowledged that these shocks existing literature related to the risks facing the
have important implications, not least for the poor tends to focus largely on rural settings,
poor, including short-term effects on consump- mainly in Africa and Asia. Part of the reason is
tion and nutrition, resulting in calls for and the that worldwide, most of the poor live in rural
establishment of safety nets or other social secu- areas and drought or flooding risks are the most
rity mechanisms. This paper goes beyond this commonly studied when considering the impact
view by arguing, first, that the costs related to of risk on the poor. The relatively higher
these risks are much higher than a simple con- urbanization rate of Latin America implies that
sideration of short-term costs, and secondly, that urban risks are crucial for our discussion. The
expanding insurance provision for the poor second section will take up these issues in more
could be an important instrument with substan- detail.
tial long-term welfare benefits. Most impor-
tantly, this paper discusses the scope and prob- The poor do not just undergo high risk in their
lems related to the expansion of insurance environment, rather they actively try to manage
mechanisms and products, with a focus on Latin it and cope with its consequences. A study of the
America, starting from a consideration of how strategies to manage and to cope with risk helps
risk affects the poor and the ways in which they understand the implications of risk for welfare as
respond to it. The paper discusses the most well as for the design of policy responses, in-
promising products, institutional setup and the cluding insurance. Much research was con-
required regulatory framework to successfully ducted in recent years on these strategies and
expand insurance for the poor. their implications, even though more work is
definitely needed. These strategies typically in-
In addressing the case for extending insurance to volve households trying to shape the risk they
the poor, a number of key questions need to be face by changing their activity and asset portfo-
answered. First, is risk prevalent, and what are lios so there is less risk involved. A typical ex-
these risks? Thinking about the design and pro- ample is diversification of activities, whereby
motion of specific insurance products requires a imperfect correlation between the return to ac-
careful understanding of the risks the poor face tivities is exploited to reduce overall exposure to
and their consequences. Recent surveys have risk. A key implication is typically that mean
highlighted the variety of risks the poor face return is forgone when moving to a less risky
(Morduch, 1995; Townsend, 1995; World Bank, portfolio, which effectively increases or per-
2000; Dercon, 2002; Fafchamps, 2003). Some of petuates poverty in the long run. Other strategies
these risks are relatively straightforward to in- involve risk-coping mechanisms, such as trying
sure (such as funerals, serious health problems, to overcome missing or imperfect credit and in-
unemployment) while several factors come into surance markets by entering into ‘self-insurance’
play in the case of others (a country-wide reces- via savings, in which assets are accumulated in
sion or crime) . Any discussion about insurance good years to be depleted in bad years, or enter-
for the poor will need to acknowledge the short- ing into informal mutual assistance arrange-
comings of an insurance-related approach and ments within families or neighbourhoods. Most
the need for alternative mechanisms to deal with evidence suggests, nevertheless, that risk man-
the implications of particular shocks. This paper agement and coping is rather imperfect, and
argues that there is a need to think in terms of shocks result in substantial fluctuations in wel-

fare outcomes, undermining also the asset base lems facing the poor. Alternative (complemen-
of households for future wealth creation, not just tary) measures are discussed as well, not least in
in terms of physical and financial assets, but also response to the realization that some economic
nutrition and human capital (Morduch, 1995; shocks or social and political risks usually can-
Dercon, 2002; Dercon and Hoddinott, 2004). not be insured by insurance markets or at least
that alternative measures may be more cost-
The evidence that risk strategies result in lower effective. Still, the scope for insurance products
long-term income and that shocks significantly for the poor remains strong.
undermine the ability to grow out of poverty has
important implications for the welfare costs in- The fifth section discusses general issues related
volved in risk. The overall result is not only to the design of insurance for the poor. A key
fluctuations in welfare levels, but also a loss of issue to be considered is that one must ensure
efficiency in that the poor are induced to use that the poor are effectively reached, which sug-
their assets less efficiently than the rich. Theo- gests the need to involve local and grass-root
retical models such as the one by Banerjee and organizations with established links with the
Newman (1993) build on this feature to show poor. Equally important is that the system
that risk may well result in poverty traps, a situa- should be able to provide a cost-effective service
tion in which those who cannot escape poverty and be sustainable. Insurance provision is a spe-
by their own means end up living in permanent cialized service, and should involve private and
poverty, even if other sectors of the economy are possibly public sector institutions with the ex-
growing. The implication is also that there is no perience and financial capacity to operate such
trade-off between equity and efficiency when schemes. A ‘partner-agent’ model is most likely
measures are taken to avoid those poverty traps. the most effective institutional arrangement. The
In other words, there is a case for providing in- sixth section then discusses possible products,
surance at subsidized rates so that some do not addressing the risks that are most suitable for
slip into poverty. The third section further ex- insurance-based protection, focusing on life,
pands on this point. health, property and weather insurance. Exam-
ples from Latin American countries such as Co-
Although public policy and interventions can lombia, Guatemala and Mexico are used to illus-
reduce risk (even if it is by means of subsidies), trate key problems and solutions related to the
this does not necessarily settle the issue about design and delivery of these products. The sev-
the appropriate form of such interventions. In- enth section centers on the role that different
deed, it still would need to be shown that insur- institutions should play, and focuses on the re-
ance is the right solution. In the fourth section quired regulatory framework. The eighth section
the case for strengthening insurance and insur- analyzes the potential role of local social institu-
ance substitutes is made, but it is also acknowl- tions already providing informal insurance,
edged that insurance products can be costly and, while the last section provides a set of conclu-
more importantly, they cannot solve all prob- sions.

Risk and the Poor

There are a number of ways of classifying risks ing the poor. Nevertheless, the focus has largely
faced by the poor. Two issues are relevant for been on data from South Asia and, more re-
our purposes: the extent to which the poor are cently, from Africa (Morduch, 1995; Dercon,
affected by these sources of risk, and the extent 2002). One key difference between these regions
to which developing more insurance is an ap- and Latin America and the Caribbean is the de-
propriate response. To discuss the first point, the gree of urbanization, more specifically, the sub-
covariance of risks across a population and the stantial urban nature of poverty. Whereas pov-
frequency of risks over time are relevant here. erty is mainly a rural phenomenon in Africa and
Insurance contracts are most easily offered if Asia, the urban share of population in Latin
risks within the relevant population are not co- America and the Caribbean is large enough to
variate—so that only some put in a claim at the ensure that urban areas account for most of the
same time. Furthermore, insurance for rare and poor. 1 Table 1 reports figures for a sample of
infrequent events is also typically more difficult countries with available data. Chile and Brazil
to offer. Taken together, if these rare events are are extreme cases where 84 percent and 70 per-
also covariate, i.e. typically occurring to large cent, respectively, of the poor belong to the ur-
population at the same time (such as a flood, ban sector.
hurricane or an economic recession), then insur-
ance contracts are most difficult to offer. These Urban poverty is different from rural poverty.
considerations are important for the rest of this For instance, there are risks that are specific to
discussion. the urban poor but not to the rural poor, and vice
versa, or at least risks that have a different inten-
It is important to bear in mind that, when look- sity than in a rural setting.2 Examples are sanita-
ing at how the poor are affected by risk, our ob- tion and public health risks related to over-
servations on which risks affect living standards crowding, or risks related to crime. Thus, an
most are largely based on evidence that takes analysis of risk and insurance for the poor in
into account the mechanisms people use to man- Latin America and the Caribbean will necessar-
age and cope with risk. For example, it may be ily diverge in some way from most previous
the case that the lack of old age security is not studies based on African or Asian experiences
quoted as a serious risk in a particular poor and focused, almost exclusively, on risks in rural
community since the community is still using areas. The focus here will not be on urban risks
intergenerational transfers as an effective only, but an attempt is made in this discussion to
mechanism to support the elderly. Furthermore, give at least equal weight to both types of set-
it may be that the development of market-based tings.
pension funds may crowd out these community-
based mechanisms. It has been shown that this Few studies have systematically tried to record
may even lead to some being more exposed to the sources of risk faced by the poor. As part
risk than before (Attanasio and Rios-Rull, 2000).
This has two important implications: first, we 1
While the incidence of poverty in rural areas
need to carefully study the way individuals, reaches 64 percent, it falls to 48 percent in urban set-
households and communities cope with risk, and tings (ECLAC, 2002). Ravallion (2001) dwells on the
secondly, problems of crowding-out and their link between urban shares in total and poor popula-
possible welfare implications should be dis- tions.
cussed in more detail. The fourth and fifth sec- It should be pointed out here that the urban poor
tions analyze these points further. face different risks, not more (or less) risks. For ex-
ample, in an LSMS in Peru in 2001, 30.1 percent of
Turning to their typology, in recent years a urban households reported to have suffered a shock.
In the rural sample, this percentage was 29.6 percent
number of studies have highlighted the risks fac- (Chacaltana, 2002).

Table 1. Percentage of the Poor Living in Urban Areas

Latin America and
Africa % Asia % the Caribbean %
Algeria (1995) 32.1 Bangladesh (2000) 14.7 Bolivia (1999) 50.5
Cameroon (1984) 70.3 Cambodia (1999) 19.4 Brazil (1999) 69.7
Chad (1996) 73.8 Fiji Islands (1990) 64.5 Chile (2000) 84.4
Egypt (1996) 49.1 India (1999) 25.8 Colombia (1999) 56.8
Ghana (1992) 32.4 Indonesia (2002) 35.0 Costa Rica (1999) 42.5
Guinea Bissau
(1991) 16.2 Kazakhstan (2002) 39.4 Dominican Rep (1997) 55.4
Kenya (1992) 18.0 Kyrgyz Rep (2000) 29.7 El Salvador (1999) 45.0
Lesotho (1993) 10.2 Lao PDR (1997) 11.9 Guatemala (1998) 30.1
Madagascar (1994) 15.7 Malaysia (1999) 25.5 Honduras (1999) 40.7
Morocco (1999) 34.1 Maldives (1998) 10.9 Mexico (2000) 47.7
Niger (1993) 17.7 Mongolia (1998) 48.8 Nicaragua (1998) 57.9
Nigeria (1993) 34.2 Myanmar (1997) 34.8 Panama (1999) 61.5
Senegal (1991) 14.3 Nepal (1996) 5.2 Paraguay (1999) 43.2
Sierra Leone (1989) 27.1 Pakistan (1998) 20.5 Peru (1999)] 48.8
Papua New Guinea
Tunisia (1990) 37.3 (1996) 5.1 Venezuela (1994) 78.5
Zambia (1993) 2.5 Philippines (2000) 29.2
Zimbabwe (1991) 10.3 Sri Lanka (1995) 8.5
Thailand (2000) 35.3
Vietnam (1998) 5.0
Sample average 31.4 Sample average 23.4 Sample average 58.4
Own calculations. Urban population shares are those implicit in urban, rural, and national poverty rates. Sample
averages are weighted by 2001 total (poor) population figures. Note that the African sample is arguably more ur-
Sources: African Development Bank (2003), Asian Development Bank (2003), and ECLAC (2002).

of a more extensive World Bank study on pov- other man-made shock, and 13 percent reported
erty in Guatemala, the analysis by Tesliuc and both. The commonly reported shocks were agri-
Lindert (2002) provides a unique insight into cultural related but many different types of
those sources and their consequences in this shocks were identified (Table 2). Some of these
country, based on a specifically designed house- shocks were largely ‘rural’—pests, lost harvest
hold survey combined with focus group inter- and drought are the obvious ones—, while oth-
views, conducted in 2000. It should be noted ers are common in urban areas—including crime
that Guatemala is one of the most rural econo- and job losses, which had more than double the
mies in Latin America, presenting the lowest incidence in urban areas. Accidents and floods
percentage of urban poor (table 1), so their find- have similar incidence in rural and urban set-
ings need to be supplemented by other sources tings, with no apparent differences in the overall
for a more complete picture. Although no seri- reported incidence of shocks across both areas.
ous economic crises or natural disasters occurred Other studies in the same period confirm the
in 2000, Tesliuc and Lindert found that about 53 high incidence of shocks. For example, Gaviria
percent of the households interviewed reported and Pages (1999) report that in the first semester
one or more shocks: 23 percent mentioned a of 2000 36 percent of urban Guatemalans re-
‘natural’ shock (from pests to forest fires and ported a shock causing loss of income. A study
floods), 17 percent reported an economic or on Peru (Chacaltana, 2002) reported about 30

Table 2. Incidence of Reported Shocks in Guatemala in 2000

Frequency Reported Type of Shocks
15 or more Pests, lost harvest
6-14 percent Income drop, accident of breadwinner, job losses, drought
2-5 percent Loss of terms of trade, crime, floods
0-2 percent Hurricane, bankruptcy, land slide, death of breadwinner, enterprise clo-
sure, land dispute, fire, earthquake
Source: Tesliuc and Lindert (2002).

percent of households facing a significant shock Tesliuc and Lindert (2002) did not analyze
to income or wealth in 2001. In short, all avail- health shocks in their survey, which does not
able evidence suggests that shocks are prevalent mean that they are irrelevant. For example, in a
in both rural and urban households in Latin rural sample in Mexico (World Bank, 1995, re-
America and the Caribbean. ported in Ibarra, 2003), it was found that about
48 percent of households reported a shock in
The study by Tesliuc and Lindert (2002) has wealth or welfare related to a drop in yields
other striking findings. First, in 2000, all these (largely due to weather-related events), while the
shocks were typically idiosyncratic, i.e. only a second main cause was illness of the farmer or a
sub-section of a local population was affected. It member of its family (reported by about 15 per-
should nevertheless be recalled that this is cent of households). This is consistent with sur-
largely a year without serious ‘large’ and covari- veys from across the world where illness is typi-
ant shocks. In fact, during the last five years, cally the second most frequent risk in rural set-
some shocks had a much higher incidence, in- tings after crop failures (see Dercon, 2002), and
cluding hurricanes (Hurricane Mitch was in- before many other risks (such as loss of live-
cluded in this period), with 44 percent reportedly stock, crime or fire).
being affected, while forest fires hit 17 percent
at least once over a five-year period. These types In sum, a discussion of the risks faced by the
of risks are largely covariant, affecting typically poor in Latin American and Caribbean countries
whole communities or neighbourhoods. Sec- will need to consider those risks with high inci-
ondly, there was a rather high incidence of dence and with serious consequences. ‘Natural’
households being affected by more multiple risks are clearly at top of the list, but not only for
shocks—a phenomenon they call risk ‘bunch- rural settings. Some are obviously rural, such as
ing’, which may exacerbate the consequences of those related to harvest losses due to drought or
shocks. Agricultural shocks, such as drought and pests, but floods or hurricane risks, and other
pests, tend to come together, as are economic large-scale natural disasters, are also relevant to
shocks, such as job losses and accidents or death urban settings. Data on the impact of large-scale
of a breadwinner. They also find that the poorest natural disasters over a 30-year period (table 3)
are more affected by shocks. In terms of asset or suggest that their impact on households—
welfare loss, the poor are typically hit the hard- ranging from death, injury, homelessness and
est. This is especially the case for shocks related physical damage—is substantial.
to agricultural risks.

Table 3 Effects of Natural Disasters in Latin American and Caribbean Countries, 1970-2001

Central South
Caribbean America America Total
Affected (000’s) 19,774 20,146 104,980 144,900
Killed 5 85 156 247
Injured 8 202 276 486
Homeless 971 2,664 4,240 7,875
Damage (US$ 000’s) 10,187,949 23,121,364 35,192,517 68,501,830
Annual damage 318,373 722,543 1,099,766 2,140,682
From Chacaltana (2002), based on CRED data.
Other risks, not least health, disability and mor- water, sewage) was ranked among the urban
tality risks, should be given center stage as well. poor as their greatest problem, while poor envi-
Health care costs cause significant stress among ronmental conditions were also important in ur-
the poor. In fact, in one of the most careful stud- ban areas in Ecuador: “Children fall into the
ies on risks faced by the poor in an urban setting, mud. The river is full of garbage. (…) People
illnesses are the most common shocks in the have no bath. All the garbage goes into the
SEWA data from India (Chen and Snodgrass, river. It is dangerous because of the tides. The
2001). In Peru, the LSMS of 2000 shows that water flows inside” (World Bank, 1999). When
health expenditure adds up to 8.9 percent of total trying to determine possible interventions to ad-
monthly expenditure of the poor when a house- dress health risks, it is impossible not to take
hold member is ill. into account these issues or consider insurance
and other forms of protection.
Illness is a more pervasive risk in areas where
public health services are inadequate. Careful A number of largely economic risks, such as job
qualitative work for the World Development loss and lower income, requires attention as
Report 2000/01 for a number of Latin American well. Qualitative surveys on perceived risks
and Caribbean countries suggests that both the stress the central importance of the wage labor
urban and the rural poor feel they have little ac- market, not least in urban areas (e.g. Zaffarino,
cess to good quality health services, although the 1999; World Bank, 1999). Other related risks
issue is mentioned more frequently by the latter are relative price changes and general inflation.
(World Bank, 1999). According to the LSMS Urban households are typically much more ex-
from Peru, 46.8 percent of the urban poor has posed to these risks, since they rely more on the
access to a doctor, while this percentage falls to market, a fact Moser (1998) calls ‘commoditiza-
39.6 percent among the rural poor. tion.’ For example, it is easier for the rural poor
to withdraw from the market during inflation
In the case of disability, illness creates addi- spells. In Peru, 13.8 percent of urban households
tional health care costs, but also a permanent considered the economic crisis as a shock in
effect due to the loss of income-earning capac- 2001, while only 2.8 percent of rural households
ity. Even a temporary disability may result in job did so (Chacaltana, 2002).
loss, and the low-income period may last until
the household member finds a new job. This is Finally, a discussion of risks faced by the poor
especially harmful to the urban poor, as unem- needs to address some crucial ‘social’ risks, in-
ployment is a greater threat to them. The death cluding crime and lack of protection and rule of
of a family member brings about grief and sig- law. One of the key findings of the qualitative
nificant economic costs. Some of them are one- studies as part of the World Development Report
off outflows (burial costs) while, more impor- 2000/01 was that crime especially affects the
tantly, others are permanent (loss of a source of poor, which is, for the most part, linked to a
income). poorer protection by the police and judiciary
Risk of illness is often closely related to particu-
lar environmental risks, such as inadequate Brazilian favelas are an evident and extreme
waste disposal, water supply, and sanitation. In a example of the threats crime imposes on the ur-
study of urban vulnerability and risk, Moser ban poor. In urban areas in Argentina, “insecu-
(1998) calls these risks environmental hazards, rity is constant and daily. (…) There is more
and considers them as one of the “three charac- insecurity (…) in the slums, because they do not
teristics of urban life often identified as differen- have material resources to face insecurity nor
tiating urban from rural areas,” along with support from the government (…) ‘The police
“commoditization, and social fragmentation.” does nothing’” (World Bank, 1999). In rural vil-
For example, in qualitative studies in Bolivia lages, “security is not mentioned [as an issue].
related to the World Development Report (…) Their perception of security is influenced
2000/01, lack of access to public services (e.g. by the news received from big urban centres:

‘They have everything but they are worse be- poor enforcement or even abuse by police or the
cause of crimes and drugs; we sleep with open judiciary system. In particular, lack of legal
doors here in the inland’” (World Bank, 1999). ownership exposes the poor to sudden losses as
Table 4 shows victimization rates in a number of the authorities force them to leave their homes,
Latin American and Caribbean countries. Al- or to plain abuses from corrupt officers. In Ar-
though there are always objections to this type gentina, the urban poor complain that ‘not only
of statistics, they show higher rates in larger cit- they [policemen] do not protect us, but they also
ies. chase us and treat us badly’ (World Bank,
1999). Similarly, theft is reported as a constant
The perceptions of these risks on the part of the threat to the assets of the poor, whose
poor are often closely linked to the absence of neighbourhoods have usually no police protec-
property rights and the rule of law because of tion.

Table 4. Victimization Rates in Some Latin American and Caribbean Cities, 1996-1998

City Size
Small Medium Large
Argentina 19.4 30.8 40.3
Bolivia - 33.9 35.5
Brazil 42.2 43.7 40.2
Colombia - 35.5 44.4
Costa Rica 35.4 45.5 -
Chile 11.6 28.6 33.2
Ecuador 40.1 45.3 62.3
El Salvador 42.8 52.2 -
Guatemala 50.3 51.5 -
Honduras 38.5 53.5 -
Mexico 29.0 43.6 53.4
Nicaragua 35.5 45.3 -
Panama 26.1 38.9 -
Paraguay 29.4 36.9 36.6
Peru 25.6 32.8 41.9
Uruguay 20.0 30.1 36.9
Venezuela 38.1 47.0 54.7
Source: Gaviria and Pages (1999), based on Latino-
barometer data.

Market Failures and Household Responses to Risk

If these risks are substantial and the conse- to birth and death certificates may make insur-
quences are as serious as suggested above, the ance contracts less attractive to the poor as well.
question is why insurance markets are not offer-
ing insurance contracts to the poor. There are a Supplying the poor with insurance implies fur-
number of reasons why this may not occur. First, ther high transaction costs. For instance, experi-
the usual information asymmetries apply. Insur- ences with microcredit suggest that the poor find
ance contracts are exposed to adverse selection it easier to deal with frequent repayment in small
(hidden information) and moral hazard (hidden instalments. 3 This suggests that payments of
action). In particular, they have been pointed out premiums may also ideally occur in small in-
as the cause for the failure of crop insurance sys- stalments, adding transaction costs to insurance
tems (e.g. Braverman and Guasch, 1986; provision.
Binswanger, 1986). Similarly, health risks are
often hard to insure in a comprehensive manner, Furthermore, it has been suggested that the poor
as are substantially covariate risks, such as natu- sometimes have difficulty in properly under-
ral disasters or economic recession. However, it standing their rights in insurance contracts.
remains to be explained why these asymmetries McCord et al. (2001) report several cases where
could be more perverse when policyholders are the poor did not file their claims after being af-
poor. In fact, they also plague contracts in more fected by an event covered by their policies. In
developed markets. other cases, some policyholders expected cover-
age beyond the scope of their contract.
Insurance providers mitigate information asym-
metry by promoting group insurance (against Finally, many of the most serious risks faced by
adverse selection) and by requiring co-payments the poor may well be covariant, and therefore
and deductibles (against moral hazard). Al- not easily insured by an emerging insurance
though insuring large groups is a feasible strat- market. The fact that a sizeable part of the popu-
egy, co-payments and deductibles may well dis- lation is dependent on agriculture, and that mac-
courage the poor from buying the product. In roeconomic instability is substantially higher in
any case, these payments and deductibles will developing rather than in richer countries (re-
probably need to be lower than the values sulting in serious covariate shocks in the econ-
needed to separate ‘good risks’ from ‘bad risks’ omy) is bound to limit the emergence of private
(the so-called second-best separating-equili- insurance focused on poorer segments of the
brium values). As contracts will still allow for population.
significant moral hazard, insurers will require
high premiums and discourage the poor. The lack of market-based insurance could in
principle have been compensated for by ‘social
A related issue is that as the poor do not usually insurance’—or public sector-based insurance
participate in the formal economy, formal insur- provision as part of broader social security pro-
ers also face enforcement problems and/or the grams. In practice, the coverage of these pro-
poor confront extra costs. For example, claiming grams for the poor is minimal in most Latin
for home insurance when there are no formal American and Caribbean countries. For exam-
titles to land or homes imposes extra verification
costs, which discourages firms from offering
contracts to the poor or makes them less attrac- In fact, Armendariz and Morduch (2000) argue that
tive. Similarly, the assets of the poor may be of theoretical literature on microcredit has exaggerated
relative low value, so the transactions costs in- the focus on joint liability and dynamic incentives,
volved in valuation would be relatively high and neglected the importance of the repayment
relative to the size of the contract. Costs related schedule. This comment is bound to be relevant for
insurance provision as well.

ple, a recent review of Guatemala’s social insur- coping strategies effectively try to smooth con-
ance system concluded that the “system pro- sumption given income fluctuations linked to
vided minimal coverage to the population, risks risk. These strategies include self-insurance, i.e.
financial crisis, faces allegations of corruption building up suitable liquid assets in good years
and is regressive” (World Bank, 2003). The In- that can be depleted during a bad year. An alter-
stituto Guatemalteco de Seguridad Social (IGSS) native strategy is to enter into informal ‘risk-
covers workers in the formal private and public sharing arrangements,’ i.e. those based on recip-
sectors only, and runs a number of programs. rocal gifts or contingent credit.
Programs analyzed, such as the accident-
maternity-sickness (IVS) program, were shown Risk management and coping strategies are al-
not only to be in deficit but also regressive in ways present in the life of the household. How-
terms of the incidence of benefits. ever, if a serious crisis occurs, households resort
to more extreme actions, survival strategies, or
The lack of formal insurance or social insurance ‘emergency’ actions to be taken when a reduc-
systems does not mean that the poor are passive tion in income is unavoidable. See table 6 for a
toward the risks they face. Much of their liveli- summary of such strategies. More information
hood is centered around ways to reduce, mitigate can be found in Dercon (2002), while the Social
and cope with risks. The poor use risk manage- Risk Management Approach is discussed in
ment and risk coping strategies to alleviate risks. Holzmann and Jorgensen (2000) and in World
Table 5 describes these strategies and their Bank (2000).
shortcomings. By risk management we mean
that they try to reduce the exposure to risk or Much of this literature was developed using data
mitigate the risk of some income sources by from Asia and Africa, but many of these re-
combining them with others. Diversification of sponses can also be found in sources for Latin
crops and other sources of income is one typical American and Caribbean countries. A few strik-
example. Other common strategies involve mi- ing conclusions emerge, which are common to
gration and relative specialization in low-risk the empirical literature on this issue. First,
activities, even at the cost of lower returns. Risk- households cope with risk by using income-

Table 5. Risk Management and Coping Strategies

Strategy Examples Shortcomings
Managing and reducing risk result- Crop diversification Forgoing expected income
ing from changes in sources of Specialization in low-risk ac-
income tivities
Migration of some members of
the household
Asset management Savings as self-insurance Lack of suitable saving assets
(risky or bulky assets, insecu-
Focus on liquid, less produc-
tive assets
Long building-up time
Covariance in asset price and
Informal insurance Reciprocal gifts/loans from Incomplete protection
friends/relatives Vulnerability to covariant risks
Market-based Insurance Typically not available

Table 6. Survival Strategies

Strategy Examples Shortcomings
Changes in sources of income Child labour Sacrifice of human capital
Asset management Selling/pawning of Long time to replace them
real/productive assets
Informal insurance Charity Incomplete protection
Vulnerability to covariant risks
Market-based Bank loans for consumption Usually not available

based strategies—such as diversification of in- fers and support was minimal in both countries;
come sources—and assets for buffering con- the most significant support went to the rural
sumption. Informal insurance and credit is also poor in Peru—but then ‘relying on State sup-
used, but only in a relatively limited number of port’ still accounted for only about 4 percent of
cases. the responses reported.4

For example, as reported in table 7, Chacaltana These strategies are not without cost. As has
(2002) suggests that most Peruvian households been widely documented, both income- and as-
(72 percent) dealt with shocks on their own, ei- set-based strategies imply efficiency losses in
ther through changes in their portfolio of income the generation of income, and thus may lead to
sources (an additional household member en- poverty traps (e.g. Rosenzweig and Binswanger,
tered the labour market), or through asset man- 1993; Rosenzweig and Wolpin, 1993; Dercon,
agement (their savings were used, or they sold 2002). 5 Lack of formal or informal insurance
off or pawned their assets). Informal insurance is forces households to choose a safe portfolio of
not widespread in Peru. Only 23 percent of activities and assets, which typically implies a
households resorted to it through loans or gifts lower mean return. The poor have no access to

Table 7. Exogenous Shocks and Household Responses in Peru, 2001 (%)

Urban Rural Total
Did something to solve the crisis 87.5 71.0 81.7
Self-help 76.1 65.3 72.3
Informal insurance 28.6 13.0 23.1
Public sector 1.0 4.3 2.2
Market-based 0.5 0.1 0.3
Other 8.8 5.7 7.7
Did nothing to solve the crisis 12.5 29.0 18.3
From Chacaltana (2002), based on ENAHO 2001-IV.

from relatives or friends. Interestingly, although
we should expect rural villages to develop Note that these are the responses obtained after a
stronger social networks, this type of insurance shock did occur. Households that successfully
is more common in urban areas. However, it is avoided shocks by directing their efforts toward those
unclear that the same pattern should be expected activities that offered more stability would not have
in other countries. Similarly, in Guatemala, Tes- faced as many shocks as those that did not, implying
that the percentage relying on income-based strate-
liuc and Lindert (2002) found that self-help ac-
gies is actually higher than data would suggest.
counted for more than half the responses about 5
how they coped with a shock, and informal in- It is worth noting that most studies focus on poverty
traps in the rural sector. Hence, the effect of risk ex-
surance via transfers only accounted for about
posure on urban investment decisions remains to be
13 percent of the responses. Government trans-

insurance, and lack of insurance precludes the to them more than the rich, such losses are es-
poor from taking risks and increasing their in- pecially borne by the poor (Rosenzweig and
come, thus perpetuating their poverty. The proc- Binswanger, 1993). It also means that the wel-
ess is exacerbated since, asset holdings are diffi- fare losses caused by lack of insurance are well
cult to rebuild after they are depleted. In emer- beyond those in terms of fluctuations and other
gency cases, households are also forced to forgo transient effects in consumption and other wel-
human capital, as is the case when children drop fare indicators. They involve permanent or
out from school and start working (Pizarro, chronic effects on poverty, implying substan-
2001). In Peru, Jacoby (1994) finds that “chil- tially higher welfare costs and lower efficiency.
dren from households with lower income (…) These efficiency losses also mean that specific
and greater childcare responsibilities begin interventions could be implemented if there is no
withdrawing from school earlier.” In urban ar- trade-off between efficiency and equity, and if,
eas, issues of privacy in the household arise, as by increasing equity (spending focused on the
families rent out rooms, or children come back poor), efficiency is increased. 6 (It implies that
to the parental home so that they can rent out schemes to promote insurance for the poor may
their own house (Zaffaroni, 1999). In fact, after well have a subsidy element that could enhance
analyzing data on Guatemala, Tesliuc and Lin- efficiency. If providing insurance would mean
dert (2002) concluded that “the poor have lower that the poor can take on more risky, but higher
resilience than the rich to the effects of shocks. return activities, then in principle, these schemes
The probability of restoring household income may be able to pay for themselves in efficiency
to the level that prevailed before the occurrence terms. This makes the case for interventions to
of the shock rises with income.” encourage insurance with public (and aid)
money for reasons that go beyond the promotion
In short, risk strategies tend to result in effi- of equity (see Dercon, 2004).
ciency losses, and since the poor have to resort

For a more detailed discussion, see Dercon (2004).
explored. This research is especially relevant for For a theoretical discussion on poverty traps induced
Latin America. by risk, see Banerjee (2004).

Scope for Insurance Provision to the Poor

The previous section analyzed the benefits of tributive transfers is that they may be able to
facilitating and enabling efforts to ensure that address many of the causes that trigger poverty
risk and its consequences are reduced for the and hardship within one system. For example,
poor. It also identified a number of risks that hardship could be linked to low assets or to a
especially affect the poor (natural, health- shock. Also, risk has a more substantial impact
related, economic and social risks). Indeed, there on the poor due to their lack of assets and re-
may be an efficiency case for government action sources to cope with shocks, which excludes
in the form of providing financing and subsidis- them from credit markets. A safety net or other
ing these efforts, beyond obvious equity argu- redistributive effort focusing on those with cur-
ments for supporting the poor. Still, this does not rently low income would not need to distinguish
address the question about the form these efforts between hardship caused by a particular shock,
should take. More specifically, is it ‘insurance’ low assets or any other form of exclusion from
that provides the answer, or should other markets.
mechanisms be considered? The lack of an in-
surance market is the underlying cause for risk- A singular focus on ‘safety nets’ has serious
induced hardship, so efforts could focus on es- problems as well. First, they may not be the
tablishing or fostering such a market. Still, this most cost-effective means for addressing the
is not the typical policy answer observed. The problem of risk. Typically, they would only of-
more traditional method for dealing with ‘risks’ fer support once an uninsured risk has already
has been to provide safety nets, systems of tar- caused serious hardship. Secondly, they are
geted interventions focused on particular groups characterized by serious problems resulting from
affected by hardship, including those produced their functioning and inclusiveness, which is
by shocks. In fact, this is typically the only op- also the case in Latin America (Lustig, 2000).
tion considered. Indeed, the discussion in the third section sug-
gests serious shortcomings in the protection of-
This focus has some justification: insurance fered at present. From the point of view of the
market failures are not easily addressed. For ex- poor, current safety nets are at best a source of
ample, if asymmetric information is the root uncertainty, and at worst, the poor are excluded
cause for the lack of private insurance markets, or support comes too late. If insurance against
there is little reason, in general, for the public risk is supposed to allow the poor to engage in
sector to resolve problems related to informa- risky activities that may, at the same time, in-
tion. Similarly, even if those problems can be crease efficiency and have a high return, then
partly resolved, transaction costs resulting from this would not be properly achieved by safety
the provision of insurance to the poor are, as nets.
already discussed, likely to be high. The admin-
istrative cost of insurance may become exces- The problems related to safety nets and broader
sively expensive, with efficiency losses that may insurance provision suggest that a complemen-
offset any gains that may be obtained as a result tary balanced approach that incorporates both
of better protection against risk. Also, large co- elements would be desirable. A more detailed
variate and catastrophic risks are unlikely to be analysis would be needed to understand the op-
easily insured, unless the development of inter- timal balance of insurance-related activities and
national reinsurance markets for catastrophic safety nets. In fact, a number of alternative poli-
risks in developing countries is fostered. Until cies should also be considered in the design of a
then, public safety net systems, financed by comprehensive system of protection against risk-
taxes and aid are likely to be more reliable and induced poverty. Broadly speaking, the system
sustainable. Furthermore, the advantage of sim- should consider ex-ante instruments and ex-post
ple safety nets in the form of targeted and redis- measures.

Ex-ante measures would provide incentives and cial assets may be risky, specially given the en-
means for the poor to protect themselves against demic risks of inflation in Latin America. Typi-
hardship: better insurance products for the poor cally, financial savings are not tailored to the
are the obvious instruments, but they should also poor, offer low or negative returns, and involve
support self-insurance through savings, and pro- extremely high transactions costs imposed on
vide access to credit in order to facilitate asset the saver. Savings should not be considered as
building and properly manage those risks that just a means to build a credit-worthy reputation
might affect income. Ex-ante measures should or mobilize capital aside from the normal eco-
also focus on reducing risk itself. Ex-post meas- nomic activity of the household. The typical
ures would provide a genuine safety net, appro- products are tailored to long-term deposits, with
priately targeted to the poor but large enough in highly punitive returns for those looking for
scale and coverage to provide broad social pro- flexible instruments to respond to unexpected
tection to assure a minimal and sustainable stan- hardship.
dard of living. Such measures could be part of a
more general welfare support system, or be spe- Credit products could also help to provide better
cifically targeted to respond to risk-related hard- protection against risk. Credit can act as an in-
ship. surance substitute, and products for this purpose
should be part of the standard portfolio of finan-
The potential role of these complementary ex- cial instruments offered to the poor. Further-
ante measures should be stressed here. A first set more, credit can help to diversify the source of
of measures involves directly reducing the risks income and build up assets. It can also increase
faced by the poor, for example, policies for basic income, reduce risk in income and enhance the
health prevention and sanitation. Better informa- ability to cope with shocks that might affect in-
tion systems on prices and weather conditions, come. Financial products for the poor should be
could have substantial benefits, while invest- flexible and take into account the fact that they
ments in technology could reduce certain types face substantial risks. Linked credit and insur-
of risk, with irrigation systems and drought- ance contracts are one option—for example,
resistant crops being very good examples. In- linking credit and health insurance. This form of
deed, this type of measures could make certain insurance is not the focus of the present report,
risks, that are presently too large or covariate to but there is definitely a need for more research
offer viable insurance for, more easily insurable on such products.
in a cost-effective way. They clearly highlight
the need for multisectoral approaches to deal As part of a general system of protection against
with risk and insurance. risk-induced poverty, there is a clear scope for
insurance targeted at the poor. In the next sec-
Other financial products can also play a role in tion, risks that can be addressed by providing
coping with risk. Savings instruments, for ex- insurance to the poor are identified in more de-
ample, have been largely undervalued as an ef- tail, which is followed by a discussion of the
fective instrument for protection against hard- type of insurance products that could be offered.
ship (Dercon, 2002; Morduch, 2004). While A number of successful experiences are pre-
credit provision to the poor has received much sented as well, focusing on strategies to deal
attention, relatively little was directed to sav- with the particular challenges of selling insur-
ings, even when they do present many advan- ance to the poor. Based on this analysis, it can
tages as an area for subsidized intervention and be argued that unsubsidized insurance for the
regulation. For example, they are not affected by poor is unlikely, the only exception being life
the information or reinsurance problems affect- insurance. The State should still have an impor-
ing credit and insurance, and transactions costs tant role to play in this regard, since it is its re-
involved in these operations, while not negligi- sponsibility to create a regulatory environment
ble, are likely to be largely restricted to the ad- that fosters insurance and financial intermedia-
ministrative handling of the savings. One of the tion. Furthermore, evidence suggests that the
key issues is that insurance by means of finan- best method for offering insurance to the poor is

the partner-agent model, in which an established examples being the crop insurance programs
insurer, possibly with public sector support, co- introduced in the early 1980s in different parts
operates with local microfinance institutions. of the world. Many of the criteria included
This points to the importance of existing infor- above apply to poor and rich insurance clients.
mal institutions as potential agents, a point that However, some of them make it particularly dif-
is discussed in the last sections of this report. ficult to profitably insure the poor. The need for
premiums to be economically affordable often
Insurance involves the pooling of risk over a means that the policy portfolio cannot actually
large number of similar units and is most appro- be covered by contributions, or that insured
priate for uncertain and high losses, which are amounts are so small that they make little differ-
greater than what a household can save for or ence to the vulnerability of the poor. SEWA, an
repay. When the loss and the degree of uncer- Indian health and life insurer, is a case in point,
tainty decrease, insurance loses out to credit and with payouts so low that they only cover about
savings. Insurance therefore involves exchang- 10 percent of losses caused by illness-related
ing the uncertainty of large losses for the cer- shocks. Insurance to the poor is traditionally
tainty of small regular payments. Policyholders fraught with high per-unit transaction costs, be-
pay for the losses incurred by others, while the cause premiums need to be small and collected
costs and risk are assumed by the insurer. For frequently, while the total amounts of policies
less uncertain or smaller losses, savings or credit are also small. Problems such as moral hazard
may be more appropriate.7 and adverse selection are not necessarily more
damaging among the poor, but the higher trans-
Brown and Churchill (2000) suggest that there is actions costs in dealing with them may mean
scope for insurance provision only when the fol- that these issues make insurance unprofitable.
lowing criteria are met: (i) a large number of Nonetheless, a number of small-size (microfi-
similar units exposed to risk; (ii) limited policy- nance) institutions, including some in Latin
holder control over the insured event; (iii) the America, already cater to the poor. Their suc-
existence of insurable interest; (iv) losses can be cessful experiences may help to develop some
identified and measured; (v) losses should not be best practice guidelines for potential entrants
catastrophic: reinsurance becomes increasingly into the small-scale insurance market who want
difficult with increasing covariance across peo- to target the poor. 8 Some of these lessons are
ple (such as a hurricane or a flood); (vi) avail- discussed in the fifth and sixth sections.
ability of historical information on a sufficiently
large number of people or property exposed to A key lesson is that ex-ante measures, in the
the same risk so that probability of loss can be form of a savings, credit and insurance system,
estimated; and (vii) premiums are affordable. may provide substantial protection to the poor,
They propose a rule of thumb by which if the but ultimately they cannot fully insure individu-
probability of a loss exceeds 40 percent, premi- als and families. In short, some ex-post measures
ums will definitely be too high to be affordable. that entail transfers to those affected by unin-
sured risk would still be necessary as part of a
There are numerous examples of insurance comprehensive system to protect the poor
schemes that have been introduced without against risk. Insurance products for the poor
meeting these criteria, one of the most infamous need to be simple, insuring only specific, highly
observable risks with measurable losses, while
high-risk groups may need to be excluded by
This feature may also explain why the poor in Latin design for the scheme to be sustainable.
America may be unwilling to purchase some of the
existing ‘formal’ insurance products available and All self-protection strategies require some outlay
instead prefer to rely on ‘autarkic solutions,’ includ- beforehand, and self-insurance fails if shocks
ing self-insurance, since the lack of appropriately
targeted and designed products would make existing
products relatively too costly for the poor, possibly The survey by Brown and Churchill (2000) pro-
outweighing the benefits. vides a number of examples.

occur in successive periods. Credit as a substi- But even if there are clear limits to insurance
tute for insurance may not be available either. provision for the poor as a solution for their vul-
Certain highly covariate and rare events are very nerability to risk, insurance is definitely an op-
difficult to insure. This means that some ‘natu- tion worth focusing on. In particular, life and
ral’ risks, such as catastrophes, may not be eas- health insurance, as well as forms of property
ily covered by a pure insurance system. Other and asset insurance, are within the possibilities,
risks require that other types of measures be ap- and even insurance against some covariate risks,
plied and market-based insurance products are such as drought or, in general, weather insur-
unlikely to be the most sensible or only re- ance. In the next few sections, a strategy to im-
sponse. ‘Social’ risks such as crime or enforce- plement such insurance schemes is discussed in
ment of property rights are examples. While it is more detail.
possible to design products that insure against
the consequences of these risks, they only ad-
dress part of the problem.

Implementing Insurance for the Poor

Some key issues pertaining to insurance man- that public policy should address, by providing a
agement need to be addressed when implement- clear institutional and regulatory environment.
ing insurance programs for the poor. This sec- This point is discussed in further detail in the
tion addresses institutional arrangements and seventh session. Within the context of a partner-
issues such as financial management, premium agent arrangement, mutual insurance funds may
calculation, distribution of services and reinsur- overcome some of the resistance against insur-
ance, with a particular focus on targeting the ance, since they mimic features of informal in-
poor. surance arrangements in which funds are often
distributed back to members at regular intervals.
It is paramount that agents involved in insurance Exiting informal arrangements may, however,
schemes have very close contact with the poor. become a part of an MFI’s established set of
This is unlikely to be achieved by either gov- procedures.
ernment agencies or by standard private insur-
ance providers. As such, institutions with close Turning to the issue of premium setting, most of
links to grassroots organizations or NGOs may the existing insurers surveyed by Brown and
be ideal agents, for example, microfinance insti- Churchill (2000) calculated their premiums ei-
tutions (MFIs), which are relatively widespread ther in house or by partnering with an estab-
within developing countries. But, since purchas- lished insurer to gain access to the required ex-
ing insurance involves a payout only in the case pertise. Brown and Churchill also found that
of an adverse shock, it is critical that insurance MFIs that cooperate with established insurers
customers clearly know and understand the are usually able to offer coverage at better
benefits they are entitled to. This requires a sim- prices. IFOCC in Peru searched for partners with
ple and clearly stated policy, swift processing of the actuarial expertise they lacked, but they were
claims and careful financial management of the unable to find an established insurer willing to
insurance portfolio by the insurance provider. To provide a product to the low-income market.
inspire trust among the clientele, adequate re- Instead, they used their own simple calculations
serves need to be held and financed through un- based on historical mortality statistics within
derwriting, reinsurance and investment. To be their credit portfolio. ASA in Bangladesh, how-
financially viable, insurers need to have a suffi- ever, followed a different and far more risky
ciently diversified investment portfolio. This is approach and based their premiums on customer
something that MFIs or other institutions work- demand, starting out with very high premiums
ing closely with the poor may often find hard to on their mandatory insurance policy. Numerous
achieve. complaints were received from their clients, so
premiums were then lowered successively until
A partner-agent arrangement, in which a local complaints stopped. While this ensures that cli-
institution or the MFI undertakes only the distri- ents are able to afford premiums and are satis-
bution of insurance services, linked with a pri- fied with the rates offered, it obviously entails a
vate or possibly public sector insurance pro- higher risk than the calculation of premiums
vider, may therefore be more appropriate when based on actuarial principles.
targeting poor customers. One of its advantages
is that it eliminates agent risk and allows the As Rutherford (1999) points out, one of the most
institutions involved to focus on their particular important demands the poor make on their fi-
strengths. It also allows local institutions and nancial services is easy access and regular small
MFIs to offer greater benefits to policyholders at payments, which impose the necessary payment
a similar cost. The most important drawback of discipline. An agency employing home service
this model is the limited availability of potential distribution and collecting premiums on a
partners. Fostering these relationships is an issue weekly basis would be well suited to the needs

of low-income households, although it may in- up markets for some of the large-scale covariant
cur high transaction costs. Integrated distribu- risks such as many natural disasters (Skees et al.,
tion, as practiced by SEWA in India, where life 2004). However, to attract reinsurance, it is of
insurance is distributed through already existing critical importance that primary insurers have
fixed deposit accounts, could help to curb these sound pricing policies and control against abuse.
costs. According to respondents in the Brown and
Churchill survey, all partner institutions in part-
Reinsurance is one element that is almost com- ner-agent arrangements and some cooperative
pletely absent in micro-insurance and similar insurers were likely to have reinsurance con-
insurance institutions focused on the poor. One tracts. However, few of the MFIs and other
of its many benefits is that it can improve the smaller organizations in their study have rein-
ability of insurers to grow, helps to stabilize fi- surance, which leaves them highly exposed to
nancial results, protects against catastrophic sudden increases in claims and prevents them
losses and improves underwriting expertise. Re- from having access to a potentially valuable
insurance in low-income markets can also open source of expertise.

Insurance Products for the Poor

This section focuses on four types of products, other risks and needs during their life cycle.
their strengths and weaknesses. In particular, it Delta Life in Bangladesh has been a pioneer in
discusses life, health, property-related and marketing this kind of product to the poor, al-
weather insurance. Life insurance is a relatively though it has experienced difficulties in manag-
low-risk product and the one most widely avail- ing its loan portfolio, potentially jeopardizing its
able to the poor. Most existing schemes offer ability to pay out the promised bonuses as the
mandatory term-life insurance as part of an out- policies mature. Such a product requires that
standing loan or savings account, thus minimiz- larger reserves be held and more sophisticated
ing their distribution costs. The majority of insti- actuarial expertise in its management. Still, the
tutions surveyed by Brown and Churchill (2000) case of Delta Life is an interesting one, and the
also limit coverage to only those policies that are promotion of such products could be worth-
not in arrears when the policyholder dies. Cer- while. But, once again, it should be stressed that
tain causes of death are also excluded, for exam- a partner-agent model with a sufficiently strong
ple, AIDS. While such life insurance for out- partner is the best option.
standing loans with simple terms works rea-
sonably well, it is important to note that it often An important question when offering insurance
protects the MFI more than the client, since to the poor is whether this can be done profita-
many MFIs would otherwise write off losses bly. All life insurers surveyed by Brown and
caused by death regardless of the availability of Churchill make profits, but it was evident that
insurance. 9 Additional benefits, such as insur- institutions with the benefit of access to actuarial
ance tied to savings rather than credit, and stand- expertise in calculating premiums appear to of-
alone term and endowment policies offer cover- fer greater value for lower premiums. Reserve
age that is more focused on the needs of the holdings differed greatly from 1.9 times the
policyholder rather than the institution. ACO- level of claims to several hundred times for very
DEP’s life saving insurance in Nicaragua is a similar policies. Ten to twenty times the annual
good example, since their policies provide a claims may be an advisable reserve holding. All
benefit that is double the amount held in savings insurers had expense ratios well below 60 per-
to the client’s beneficiaries. In Venezuela, CO- cent (claims expenses + operating costs / annual
OPERAR’s basic product provides a benefit premium revenues). For example, IFOCC
equal to the amount held in savings with the op- (Peru), used 44 percent of its premium income to
tion of increasing the coverage to double the cover claims and just 5 percent to cover operat-
amount for an increased premium. In this sense, ing expenses (due to integration within its credit
and although their availability is limited, these operations). Thus it seems reasonable that vari-
are better options for the promotion and devel- ous forms of outstanding balance insurance can
opment of insurance for the poor. be profitable in low-income communities. The
following table, taken from Brown and Churchill
In theory, endowment life insurance can provide (2000) gives some preliminary performance
low-income households with complete protec- guidelines for institutions offering life insurance.
tion against death risks and, through a saving
and loan component, partial insurance against Insuring health risks poses different and more
complex challenges for providers than offering
life insurance outright. Insurance of health risks
It is in fact plausible that if MFIs used reinsurance may suffer from adverse selection and moral
properly, the cost of mandatory life insurance linked hazard and usually entails the provision of health
to outstanding loans would be much lower for cus- care. To avoid moral hazard and adverse selec-
tomers than it is at present. In many ways this form tion, various mechanisms are in use. For exam-
of insurance is a cost that derives from inefficiency,
and not a benefit for customers.
ple, two Ugandan health insurance institutions,

Table 8. Guidelines for Institutions Offering Life Insurance

Claims Ratio Distribution Costs Reserves Claim-Processing Time
Preferred < 60% of annual < 10% of annual > twice the an- < 10 days
Range premiums premiums nual claims level

UHC and FINCA Uganda, require that more insurers implement mandatory reference systems
than 60 percent of the members of a group agree that encourage patients to use the lowest cost
to enroll before coverage is extended to a coop- treatment facility first. Few of the institutions
erative, trade union, MFI, or to a village bank. surveyed required formal underwriting before a
COHI Benin charges a small initiation fee to family purchases a policy. But some providers
new members and has a one-month waiting pe- have not designed their schemes properly. The
riod after receiving the first premium before Asociación por Salud de Barillas (ASSABA) in
policyholders can receive health care coverage. Guatemala (see box 1) did not manage to en-
To control for escalating treatment costs, some force the requirement that all members of a fam-

Box 1. ASSABA, Guatemala

In poor rural communities, access to basic health care is often severely limited, a problem that is being addressed
by community initiatives to generate health care financing through voluntary prepayment schemes. The example
of the Asociación por Salud de Barillas (ASSABA) in Guatemala (Ron, 1999) helps to shed light on this issue.
Ron compared the ASSABA experience with a study of a similar scheme (the ORT Health Plus Scheme, OHPS)
in the Philippines that was more successful.

ASSABA started a community health-financing scheme in 1994 following suggestions from the World Health
Organization. Preliminary estimates were made on current costs and out-of-pocket health care expenditures in
public and private facilities. The concept involved the identification of a contribution level that would be afford-
able to the vast majority of families, as opposed to a contribution that would cover all the costs of an optimal
benefit package. Donor funding was then mobilized to cover start-up costs. Limitations on benefits were im-
posed, and particular illnesses not considered ‘emergencies’ were excluded.

ASSABA, as a grass-roots, participatory community association created to improve the health of its members,
was better organized when compared to similar schemes in other places, such as the Philippines. However, at the
design stage, ASSABA was not yet established as an administrative body. By the time it was finally registered as
a legal entity, local conflicts made progress difficult, since the local Catholic Church—also a health care pro-
vider—contested the capitation contract of ASSABA with a hospital sponsored by the Protestant Church in the
United States. Furthermore, ASSABA, was attempting to provide community health insurance before the na-
tional authorities in Guatemala had come out with a clear policy. Although ASSABA implied that all members
of a family needed to register, this was not stated explicitly in the registration rules, exposing the ASSABA
scheme to adverse selection. The design of the benefits package also posed a serious disincentive to potential
members. In contrast to the original design, inpatient care was limited to three days. Additional charges would
fall on the patients, who in the majority of cases would not be able to afford them. This clearly contributed to the
failure. Successful schemes strictly enforced their registration rules and did not allow individual family members
to register, but also did not change benefits.

One of the lessons that can be drawn from the ASSABA experience is that the regulatory framework needs to be
firmly established, and that potential problems at the local level should be taken into account before setting up a
scheme. The benefits package should be designed with the needs of potential members in mind, while contribu-
tions should be kept sufficiently low. Rules for registration (group or family membership) need to be strictly
enforced, while a minimum period of membership prior to service can also be helpful in protecting against ad-
verse selection.

Source: Ron (1999).

ily be enrolled, so some enrolled only those Unfortunately, health insurance schemes tar-
members most likely to be ill, with a high nega- geted at relatively low-income customers are
tive impact on the sustainability of the scheme. often liable to serious losses. For their survey,
It did not impose any waiting time before poli- Brown and Churchill (2000) had access to finan-
cyholders could benefit from in-patient care. cial information for four providers. They found
that three out of those four providers were not
To provide the right incentives to the service covering their costs. One had expense ratios of
provider, various payment mechanisms can be 216 percent, which reflected the challenge of
used. When good quality control is in place, one profitably offering health care insurance to low-
form of payment that works well is capitation income households. Only COHI (Benin) was
payment: the insurance scheme pays the pro- able to avoid losses, since it only offers very
vider a fixed amount per member and the pro- limited coverage with many exclusions and re-
vider agrees to provide care, as defined in the strictions.
policy, for any member who needs it during the
period. By paying for the number of people in- Another product is property insurance—
stead of the number of services offered, the including against fire or theft. Few providers
scheme reduces the provider’s incentive to pro- have experience with this type of insurance,
vide more, possibly unnecessary services. ASS- which is unfortunate as property loss is a big risk
ABA in Guatemala used this scheme. However, especially for the urban poor in Latin America.
it also places risk solely on the provider if there A key issue is likely to be insufficiently defined
is excessive usage and the provider is unwilling property rights and titles that are difficult to en-
to agree to such a scheme. Therefore, fee for force. If assets are identifiable, their value is
service may be more practical. Alternatives are likely to be relatively low, which would increase
fixed cash subsidies given to each member to valuation costs relative to the value of the pol-
pay for health expenses regardless of actual icy. The experience of La Equidad in Colombia
claims. may nonetheless be helpful when designing new
products. La Equidad offers comprehensive cov-
Three different methodologies are employed to erage on many types of risks after conducting
fund health-care services provided for by their substantial market research about the needs of its
policies: salaried service provision, dedicated clients. As a consequence, its property insurance
health-care facilities and indemnity coverage. is not tied to an outstanding loan. Policyholders
Naturally, there are strengths and weaknesses to themselves determine the value of the asset.
each of these approaches. A system used by Since the premium is tied to the insured value of
ASABA, salaried service provision, whereby the asset, policyholders have an incentive to
health services are provided by staff working state the true value. This mechanism simplifies
exclusively for the health plan, often provides the sales process greatly. La Equidad determines
the most convenient access. However, many premiums according to the risk exposure of their
types of health care services cannot be provided clients and their type of business: service, trade
in a cost-effective manner by salaried personnel related or manufacturing.
serving only members of a certain health plan. If
they are provided locally, dedicated health-care In the case of property insurance for the poor,
facilities can offer convenient quality care, but premiums are not adjusted according to the pre-
this approach requires a higher degree of ad- ventive measures that might be in place, most
ministration to monitor the services provided likely because of obvious enforcement prob-
and how members use services covered by the lems. Instead, La Equidad offers regular group
policy. Indemnity coverage reduces administra- meetings for policyholders to train them about
tive costs for the health plan, but gives it less basic preventive measures they could take. To
control over the quality of care. Also, it may not prevent against moral hazard, two mechanisms
provide effective coverage for members who are used: deductibles and claims inspectors.
cannot afford to pay for services up front and However, all insurers indicated that sending in-
receive reimbursement later. spectors was too expensive for small claims.

There is little information on the financial per- ratio (the ratio of payouts relative to premiums)
formance of property insurance since it is so has been rather high—often above 80 percent.
rare. Also, and more than with other types of AGROASEMEX, a government-owned insur-
insurance, low-income households appear to be ance and reinsurance company, was in charge of
slow to embrace the idea of purchasing insur- managing the system. Until 2001 it provided
ance on their valuable assets. This may be be- direct insurance to farmers, although it now fo-
cause property risk is less certain, in contrast to cuses largely on providing reinsurance to fondos.
death or the health problems that clients may A fondo is a group of farmers in a more or less
have to deal with eventually. homogenous area, providing mutual insurance to
each other. An important portion of insurance
Another insurance product, which is not often and reinsurance is effectively linked to credit
offered but is receiving much attention, is— operations as well. But its relative success is
weather insurance (see, for example, Skees et largely due to its focus on the highly productive
al., 2004). Risks in agriculture that relate to and financially viable sector of large-scale
drought or other weather events remain some of commercial agriculture, as its key mechanism to
the most important ones facing the poor. In gen- save on costs on transactions and monitoring of
eral, systems to insure crops have generated high moral hazard and adverse selection. The system
costs and have, however, failed. So there is a is, though, not suitable for providing insurance
continued interest in finding alternative insur- to the poor. For small-scale and poor farmers,
ance mechanisms. Many factors are responsible FONDEN is the only available scheme, which is
for those failures, but moral hazard and the high a simple disaster-relief scheme functioning as a
costs of the loss verification process after spe- safety net since these relatively small farmers do
cific weather events are key in this regard. The not have access to credit or formal insurance
high covariance involved in agricultural risks (Ibarra, 2003).
and reinsurance compound these problems.
It has been suggested that weather-indexed
However, there have been innovative theories bonds could encourage agricultural insurance
for the design of systems that may not be so li- systems to start offering services to poor and
able to some of these problems. The idea is to small farmers. In this case, problems of asym-
supply insurance based not on the assessment of metric information are largely resolved and, for
crop losses, but on weather indexes. Given the example, insurance could be based on small mu-
observability of weather indexes, such a system tual insurance groups that can obtain rainfall
could avoid moral hazard and adverse selection reinsurance through these bonds. The bonds
issues, and in general save on transactions costs. could be priced for reinsurance, since historical
The recent evolution of international markets for data on rainfall are available and they could
unusual and catastrophic risk suggests that rein- even be traded internationally. Even without
surance by international markets may become reinsurance by international markets, there is
feasible (Skees et al., 2004). still a case for reinsurance through governmental
budget and assistance: the fact that an instrument
Experiences in Mexico with agricultural insur- may be available that can be provided at low
ance over the last few decades may help to illus- transaction costs to poor farmers supports the
trate the potential and possible drawbacks of idea of subsidizing for equity, and possibly effi-
weather insurance to provide protection to the ciency, reasons.
poor whose livelihood depends on agricultural
activities. In this country, there have been vari- One should nevertheless be careful not to ideal-
ous systems of agricultural insurance since the ize the benefits of rainfall insurance. To reach
1940s. Most have been largely unsustainable due the poor, substantial transactions costs will
to the recurring serious financial problems that surely be involved, while, to be effective, the
caused their collapse. The system in place since correlation between rainfall measured at reliable
1990 has been more stable, providing cost- stations and local yields will need to be high.
effective crop insurance. For example, the loss The latter is not necessarily guaranteed since

rainfall stations are not very common in agricul- probability of the scheme failing), making insur-
tural areas with low potential and limited com- ance more expensive for the poor. Overall, how-
mercial farming interests. Also, the sustainabil- ever, such new products deserve experimenta-
ity of the scheme will depend on the relative tion and further analysis to understand the possi-
predictability of certain events. While they affect bilities for effective delivery of weather-related
pricing, some weather phenomena, such as insurance to the poor. Indeed, programs being
global warming or El Niño, are not well under- implemented in Mexico for the use of weather-
stood. The high covariance involved will require based indexes will provide helpful insight, as
premiums with high frontloading (adding an ex- will current experimentation in other countries,
tra sum to the premium to handle the non-zero such as India (Skees et al., 2004).10

Price insurance is not explicitly considered in this
report, even though forms of price stabilization have
often been implemented for their insurance value,
while futures contracts effectively provide insurance
to farmers in an increasing number of countries. Pure
price insurance schemes are less common, but they
might yield good results if they are properly de-
signed. Since price shocks are highly covariate some
of the issues related to designing and delivering price
insurance are similar to those related to weather in-
surance. Collier (2004) discusses the possibility of
price insurance offered to producers of internation-
ally traded commodities, whereby, given its private
and social (growth) benefits, donors could underwrite
this insurance and subsidize administrative costs.
Collier argues that the benefits of such price insur-
ance schemes may well be larger than weather or
other quantity insurance schemes. Experimentation
with these types of schemes would be highly benefi-

Regulation for Insurance Provision to the Poor

The previous section analyzed a number of surers. The strategy is unlikely to involve large-
products that could be successfully offered to the scale subsidies, but government spending will
poor. The partner-agent model was identified as need to be directed toward establishing the nec-
the best mechanism for these products to effec- essary infrastructure, institutions and regulatory
tively reach the poor. This model takes advan- environment to promote this segment of the
tage of the strengths of the different parties in- market. Unfortunately, such a policy environ-
volved in insurance provision to the poor. The ment does not exist in Latin America, and, in
‘partner’ is an established insurer, with experi- fact, some of the existing regulations actually
ence and interest in broadening its insurance present a bias against the use of finance and in-
portfolio to include products suitable for the surance products targeted to the poor.
poor. In order for it to be a successful operation,
it will need to design contracts that provide the In a report by Jansson and Wenner (1997), the
appropriate incentives for the insurance ar- following regulatory requirements have been
rangements to be sustainable, while also being identified as biased against small-scale (micro)
credible to agents and its clients. Earlier, the ar- insurance providers in Latin America: high capi-
gument was made that the problems related to tal requirements, high capital adequacy stan-
risk-induced poverty traps implied a preference dards, ownership restrictions and the require-
for subsidized insurance, efficiency being the ment of new financial institutions to be capital-
main argument. This does not mean that its im- ized by cash contributions.
plementation is straightforward: the case for
subsidies or other government intervention Capital requirements in Latin America are often
opens in itself opportunities for rent-seeking on prohibitively large for MFIs, although actual
the part of the ‘partner’ toward the government, requirements differ widely. Even if the required
not least given the political economy conse- sum could be raised, few MFIs would be able to
quences of a scheme focused on service provi- gather a client base large enough to fully lever-
sion to the poor. The ‘agent’ will need to be pro- age their capital. In Colombia, regulated insurers
vided with the appropriate incentives to maintain are required to maintain a minimum investment
the sustainability of the portfolio. These agents of US$3.2 million as well as additional paid-in
are likely to be financial institutions with close capital based on the size of their insured portfo-
contacts with the lower-income segments of the lios. Recent surveys among insurance execu-
market. Microfinance institutions have been es- tives, including some in Colombia, revealed that
tablished with this purpose, although their in- they did not serve the low-income market be-
volvement in the insurance segment has been cause they did not believe they could achieve the
limited. Existing microfinance institutions could volume of business required to earn a sufficient
be encouraged to branch out into more wide- return on their investment (Brown and Churchill,
spread insurance, or assistance could be given to 2000). As governments increase minimum capi-
certain microinsurance providers for them to tal requirements over time to maintain a finan-
enhance their products. cially stable insurance industry, insurers with
healthy finances and serving low-income mar-
There is substantial scope for the government to kets can be chased out of business. In Bolivia,
effectively support the insurance market serving recent government demands for insurers to in-
the poor. A favorable policy environment can crease minimum capital led to the dissolution of
support the proliferation of insurance services Crucena, an insurer that served low-income Bo-
among the poor by facilitating the establishment livians for 24 years and that, in 1997, had a pre-
of local (micro) finance institutions, and making tax profit of US$640,000. To get around this,
insurance provision to the low-income segment some Latin American countries have created
of the market more attractive to established in- different institutional forms for MFIs, but these

are often severely limited in the type of activities Honduras, for example, institutional ownership
they are allowed to undertake. Alternatively, is not permitted, so NGOs cannot own MFIs.
low-income insurers sometimes offer insurance,
as member benefits, through cooperatives or It is clear from this discussion that existing regu-
credit unions, which are financed through inter- lation hampers the provision of financial ser-
est payments on outstanding loans. However, the vices to the poor. Often it does not even achieve
risk is that such institutions may no longer have the desired result of ensuring the financial stabil-
any external requirements for maintaining finan- ity of MFIs, but instead forces them to circum-
cial integrity. Regulation in the form of capital vent regulation and avoid external portfolio au-
requirements is required to ensure sustainability diting. It may therefore be advisable to adapt
but, in its current state, does not take into ac- some of the existing regulation, for example, by
count the specific needs and problems of micro- lowering capital requirements for microinsurers,
finance institutions. Alternative arrangements, allowing NGOs to be owners of MFIs, and loos-
such as agent-partner relationships with estab- ening the requirement to be capitalized fully by
lished insurers providing reinsurance to MFI’s cash contributions.
portfolio, could result in the same degree of sus-
tainability and protection. This is not to suggest that the accountability and
financial health of MFIs should in any way be
A second issue that is important for local MFIs compromised, but rather that regulation needs to
when trying to enter the market and forming a take account of the different needs of MFIs and
new financial institution is the requirement of their customer base. But substantial financial
capitalization by cash contributions. This is an regulation makes sense when viewed in the con-
obstacle, since MFIs are usually established by text of providing stability and credibility to the
NGOs with existing loan portfolios, and insur- entire financial system, even if it appears to go
ance is in the first instance offered as part of ex- against the needs of a niche in the system. While
isting credit relations. In these cases, NGOs are some efforts to adapt regulation to the circum-
required to transfer cash and clients to the new stances of the MFIs are necessary, lifting all
institution, which in turn is required to repay these regulations would not be advisable, even if
individual loans to the NGOs making the crea- MFIs typically would consider these rules as
tion of an MFI extremely expensive. Jansson being against their interests. The same degree of
and Wenner (1997) suggest that a possible way sustainability and credibility for microinsurers
to facilitate the set-up of a new MFI would be to could be achieved by relaxing some of the rules
allow NGOs to use the net present value of the for MFIs, in combination with incentives and
existing loan portfolio to capitalize the new in- possibly requirements for MFIs to foster links
stitution, as long as this value is adequately and with established insurers as part of partner-agent
independently evaluated. institutional models. In any case, even in the
current regulatory climate, careful use of the
Restrictions on ownership of financial institu- partner-agent model could provide a solution for
tions can also be an important obstacle to the MFIs to properly expand their activities by fol-
creation of regulated microfinance entities. In lowing the requirements imposed by the regula-

Local Institutions and Insurance Provision to the Poor

One should however be careful not to idealize agents may crowd out any local ‘informal’ in-
the ability of MFIs to easily and effectively pro- surance or other beneficial interactions. These
vide insurance to the poor. While they may be a incentives need to be discussed, focusing on the
crucial intermediary for established insurers to overall benefits of the scheme.11 Note that these
enter the low-income segments of the market, concerns could be present even if the model is
their own ability to effectively reach the poor simply an agent with close contact in the local
should not be taken for granted. Their record of community (the MFI) directly trying to intro-
reaching the poor is not always impressive. duce formal insurance on behalf of the estab-
Formal institutions have typically difficulties in lished insurer, although in the discussion below
reaching poor communities and individuals, who the focus is on the existence of informal but ex-
end up being largely dependent on their own plicit interactions between people at the local
risk-coping strategies, even if seemingly appro- level, such as in the form of informal risk-
priate alternatives are available. Any program sharing.
aimed at including the poor should be sensitive
to these problems. Several studies examine the possible interaction
of explicit incentives—those that can be exter-
One route to consider would be trying to mobi- nally verified and thus become the basis of a
lize existing ‘informal’ savings and insurance contract—and implicit incentives in principal-
institutions to assist in ‘crowding in’ financial agent contracts. This can help to understand how
services, including insurance, into these com- credit contracts could be designed in the pres-
munities. A plethora of local informal institu- ence of local informal risk sharing. Conning and
tions run by their members such as Rotating Kevane (2004) discuss the case of obtaining a
Savings and Credit Associations (ROSCAs) and loan to undertake a risky project whose suc-
Accumulating Savings and Credit Associations cess—observed by the financial institution—
(ASCAs) exist, which provide an opportunity for depends on the amount of effort (unobservable
credit, saving and insurance, while the existence to the outside financial institution) that is exerted
of more ‘informal’ groups—such as mutual sup- by the borrower. As with all problems of moral
port networks and funeral associations—have hazard, any contract that is to implement dili-
been identified throughout the world. The key gence must offer the agent a higher expected
issue is whether they can be integrated into more utility under project success than under failure,
formal insurance projects as potential local so as to give him an incentive to want to raise
agents in a partner-agent framework with the the probability of success through diligence. For
following hierarchical structure: an established this to be the case, the villager, who is assumed
insurer that contracts a microfinance institution, to be risk-averse, must be made to bear the risk.
which in turn involves a local ‘informal’ institu- The feasibility of such a contract between the
tion dealing with the clients. The key advantages villager and the financial institution thus de-
of mobilizing these local informal institutions
are their expertise, reputation and informational 11
It is sometimes argued that crowding-out of local
advantage on the local community. institutions, such as credit and insurance systems,
should be avoided at all cost. However, as Morduch
It is worth to carefully discuss how this may (1999) argued, the key issue is the overall welfare
work. Two points are crucial here. First, it may benefit of the scheme: the benefits to individuals
be that offering insurance or other products from from ‘formal’ financial intermediation should out-
outside the local community can be done more weigh the costs of the disappearance of informal
effectively using these local institutions, result- mechanisms. Note however that there may be distri-
ing in net benefits to the community. However, butional effects as well (other people benefit from
those that lose), which require careful consideration,
it may also be that the introduction of outside not least if the poor suffer more.

pends on the cost of diligence. Conning and ding formal financial services, including insur-
Kevane show that if agents in the village have ance.
the ability to enter into side-contracts for the
purpose of mutual insurance which are based on While the above has focused on bilateral risk-
observed effort, the set of feasible contracts is sharing, local informal institutions where they
increased since the side-contract can provide exist usually comprise larger groups (albeit
more risk-smoothing to the agent who took out rarely the whole community) and often hold
the loan without disrupting incentives because substantial amounts of assets.12 Genicot and Ray
monitoring by other agents keeps the latter dili- (2003) in fact show that these informal groups
gent in circumstances where the individual in- will always be of limited size because of the re-
centive compatibility constraint would not be quirement for self-enforcing arrangements in the
satisfied. This is an example where local infor- absence of legally binding contracts. This opens
mal insurance can crowd in outside financial up the possibility of offering reinsurance to such
intermediation. This rests on the assumption that groups. This would have a direct beneficial ef-
monitoring is costless, that local agents have fect of reducing the claim variance an informal
better information than outside agents and, most institution faces and, since the size of such
importantly, that they do enter into an insurance groups is constrained by the possibility of devia-
side-contract. tion during periods of illiquidity, an indirect
benefit may be that larger groups achieve stabil-
Literature on self-enforcing contracts however ity thus increasing diversification against risk
shows that this need not be the case. Because of within the informal institution. An added benefit
the nature of informal insurance, contracts need of offering reinsurance to existing groups is that
to be self-enforcing, which requires that the one- it does not change the payoffs in autarky but
time gain from deviation is smaller than the ex- only affects those of remaining in the group, so
pected benefit of continuing in the arrangement. that there is strict complementarity between in-
This means that informal insurance is not always formal and formal insurance. Furthermore, funds
feasible and furthermore its feasibility is af- of such informal institutions may be used as col-
fected by the pay-off from reneging on the lateral to crowd in loans from outside financial
agreement, which in turn is affected by the institutions, as suggested by Conning and
availability of outside financial intermediation. Kevane (2004). In short, these theoretical argu-
However, even if the pay-off from reneging is ments suggest a number of avenues in which
increased through the access to outside credit, extending financial services such as formal in-
this may not necessarily lead to a breakdown of surance to local communities and through local
existing informal insurance arrangements. Intro- institutions could have substantial benefits. But
ducing an outside safety net or other form of all these models include restrictions on the type
insurance that is well targeted not only increases of contracts and arrangements between MFIs
utility in autarky but it also affects the distribu- and the local community that are in fact possi-
tion of wealth in a community, hopefully mak- bilities for improving welfare. To put it simply,
ing it more equal and facilitating reciprocal schemes may still result in overall negative wel-
transfers, where before income differences fare effects. Furthermore, if there are different
would have been too large to make risk-sharing levels of wealth among villagers, the benefits
possible (Coate and Ravallion, 1993). Even bet- and costs may well be borne by different people,
ter results can be achieved when the availability adding further complexity to the evaluation. In
of outside financial services is made conditional any case, it points to the need for a careful de-
on participation in a local informal risk-sharing sign of insurance products and their delivery that
arrangement (Attanasio and Rios-Rull, 2000). In should take account of the functioning of exist-
other words, this implies that there could be ing local mechanisms. The analysis above sug-
ways to increase informal risk sharing by exten- gests that sensible directions for integrating local

Burial insurance, which is a simple type of life
insurance, is often organized in this manner.

informal schemes into broader insurance provi- their funds as collateral for loans, while also
sion to the poor could include offering group making use of their local expertise in reducing
policies or reinsurance to existing groups, using transaction costs and asymmetry of information.


The poor in Latin America face substantial risk, lished insurer (the partner, from the private sec-
in the form of natural, health, social and eco- tor, possibly in partnership with the public sec-
nomic risks, and are also more likely to be af- tor) links up with an institution with local finan-
fected by them. The high degree of urbanization cial connections, such as a microfinance institu-
and commoditization in the countries makes tion. The advantage is that this would include a
Latin America different from other developing mechanism to provide easy access and terms to
regions. In general, the poor use sophisticated the poor, while costs are reduced and sustain-
mechanisms to cope with this risk, which is not ability is protected by reinsurance and contract-
enough. The welfare losses are substantial: the ing with an established insurer. Judging from
coping mechanisms themselves come at an addi- case studies, it appears that a number of products
tional cost in terms of long-term welfare. As might be suitable for promotion, including life,
such, risk and how the poor respond to it con- property, health, weather or price insurance,
tribute to the persistence of poverty. There is a possibly linked with credit. By its covariate na-
clear need for further policy work to reduce risk ture, weather or price insurance requires mecha-
and its consequences, as current systems do not nisms of reinsurance either internationally or
provide sufficient protection. Indeed, there may through budget or aid, but much progress has
well be an efficiency argument for providing been made in recent years to develop workable
subsidized insurance and protection, given risk- models.
induced poverty traps.
One should be cautious about the likely success
This study has argued for fostering insurance of these schemes. In terms of types of coverage,
provision, not as panacea to solve all problems, the experience of existing insurers catering to
but as part of a comprehensive system. The cur- the poor shows that it is difficult to offer profit-
rent focus on ex-post measures in the form of able comprehensive coverage to low-income
some safety net is not cost-effective or sufficient households. In part, this can be explained by the
to reach the poor. Other components of such a financial capacity of clients and the lack of op-
system would be ex-ante measures to stimulate portunities for diversification. Term life insur-
and protect self-insurance through savings, re- ance is the most sustainable type of insurance,
ducing risk and fostering credit for the poor as a but the support of governments, donors and
form of insurance and to allow a stronger asset NGOs is necessary to branch out into other prof-
base to grow out of persistent vulnerability to itable products. Product features should include
risk. These efforts need to be supplemented by a group policies, mandatory insurance and incen-
careful and well-designed safety net since some tives to cope with moral hazard and adverse se-
risks should not be addressed by ‘ex-ante’ insur- lection, for example by rewarding members who
ance-related mechanisms—examples are certain do not submit any claim during the year.
covariate economic or catastrophic risks. A high
proportion of risk, including economic and so- As an indirect benefit, the provision of sustain-
cial risk, is also largely man-made, and reducing able insurance services creates natural incentives
their impact requires actions to address the for insurance companies to encourage risk pre-
causes of these risks—inflation, crime or waste- vention as in the case of La Equidad, in Colom-
related risks are examples. Providing only pro- bia. Many existing schemes proved costly but
tection against the consequences of these risks is also clearly lacked expertise and reinsurance
unlikely to be cost-effective. mechanisms to reduce costs. The partner-agent
model is therefore likely to be the most efficient
In terms of the basic institutional setup for in- way to proceed.
surance provision, the partner-agent model ap-
pears to be the most suitable, so that an estab-

While subsidized insurance for the poor can be tivities. In short, it would allow the poor to focus
an attractive option on efficiency grounds, an on long-term strategies to get out of poverty
important role for the government would be to
establish a more effective regulatory framework Of critical importance are the credibility and
to foster the establishment of microinsurers at sustainability of insurance provision as part of a
the local level, while maintaining overall stabil- broader social protection system. The issue is
ity and credibility of the entire financial system. not who should provide the services as part of
While relaxing entry requirements for MFIs to the system: different agents, including NGOs,
enter the insurance market may be beneficial, community organizations or the private sector
incentives should be also provided so that MFIs could play a significant role. There is a key role,
partner with established insurers, through the however, for the government in the development
partner-agent modality. of and support to an appropriate regulatory and
institutional framework for such programs, and
MFIs can easily and effectively provide insur- sustainable and transparent institutions to moni-
ance services to the poor, but it is important to tor these activities.
acknowledge the presence of local insurance and
other finance-related institutions. There is a clear This issue cannot be underestimated. Often, in-
scope for involving these institutions as inter- stitutions in developing countries, including
mediaries in insurance provision, since they pos- those in Latin American and the Caribbean, are
sess knowledge of local conditions and an estab- not transparent or sustainable, and therefore
lished reputation. well-intentioned measures may lack the credibil-
ity and public support to succeed. Credibility
Finally, objectives should be clearly defined cannot be easily acquired, and governments face
when providing more insurance to the poor. Un- an uphill struggle in this regard. The important
insured risk means that poverty is perpetuated, role for aid and the donor community is to sup-
with the possibility that a risk-induced poverty port and ensure the enforcement of these meas-
trap might occur. More insurance, as part of a ures so that the benefits of insurance products
credible comprehensive system of ‘social protec- targeted to the poor can be improved by enhanc-
tion,’ should allow the poor to sustain their as- ing their long-term effect on poverty.
sets and to enter into more profitable, risky ac-


African Development Bank. 2003. Gender, Poverty and Environmental Indicators on African Countries,

Asian Development Bank. 2003. Key Indicators 2003: Education for Global Participation (Part III: Re-
gional Tables).

Attanasio, O. and J. Rios-Rull. 2000. Consumption Smoothing and Extended Families. Mimeo.
University of Pennsylvania.

Banerjee, Abhijit and Andrew Newman. 1993. Occupational Choice and the Process of Development.
Journal of Political Economy, 101, 2: 274-98.

Banerjee, Abhijit. 2004. The Two Poverties. In Insurance against Poverty, ed. S. Dercon. Oxford Univer-
sity Press.

Besley, Timothy. 1995. Nonmarket Institutions for Credit and Risk-Sharing in Low-Income Countries.
Journal of Economic Perspectives, 9(3), pp. 115-127.

Binswanger, Hans. 1986. Risk Aversion, Collateral Requirements, and the Markets for Credit and Insur-
ance in Rural Areas. In Crop Insurance for Agricultural Development, eds. P. Hazell, C. Pomera-
da and A. Valdes. John Hopkins University.

Braverman, Avishay, and J. Luis Guasch. 1986. Rural Credit Markets and Institutions in Developing
Countries: Lessons for Policy Analysis from Practice and Modern Theory. World Development,
14 (10-11), pp.1253-1267.

Brown, Warren, and Craig Churchill. 2000. Providing Insurance to Low-Income Households. Part I:
Primer on Insurance Principles and Products. Microenterprise Best Practices.

Brown, Warren, and Craig Churchill. 2000a. Insurance Provision to Low-Income Households. Part 2: Ini-
tial Lessons from Micro-insurance Experiments for the Poor. Microenterprise Best Practices.

Brown, Warren, and Craig Churchill. 2000b. A Cautionary Note on Microfinance Institutions and Donors
Considering Developing Microinsurance Products. Microenterprise Best Practices.

Chacaltana, Juan. 2002. Social Funds and the Challenge of Social Protection for the Poor in Latin Amer-
ica. Conference on Social Protection, ADB-IADB.

Chen, Martha, and Donald Snodgrass. 2001. Managing resources, Activities, and Risks in Urban India:
The Impact of SEWA Bank. USAID. Assessing the Impact of Microenterprise Services Project.

Coate, Stephen and Martin Ravallion. 1993. Reciprocity without Commitment: Characterisation and
Performance of Informal Insurance Arrangements. Journal of Development Economics, Vol. 40,

Conning, Jonathan and Michael Kevane. 2004. Why isn't there more Financial Intermediation in Develop-
ing Countries? In Insurance against Poverty, ed. S. Dercon. Oxford University Press.

Collier, Paul. 2004. The Macroeconomic Repercussions of Agricultural Shocks and their Implications for
Insurance. In Insurance against Poverty, ed. S. Dercon. Oxford University Press.

Dercon, Stefan. 1996. Risk, Crop Choice, and Savings: Evidence from Tanzania. Economic Development
and Cultural Change, 44(3), pp. 485-514.

Dercon, Stefan. 2002. Income Risks, Coping Strategies, and Safety Nets. World Bank Research Observer,
17 (2), pp. 141-66.

Dercon, Stefan and John Hoddinott. 2004. Health, Shocks and Poverty Persistence. In Insurance against
Poverty, ed. S. Dercon. Oxford University Press.

Dercon, Stefan (ed.). 2004. Insurance against Poverty. Oxford University Press.

Economic Commission for Latin America and the Caribbean (ECLAC). 2002. Panorama social de Amé-
rica Latina, 2001-2002.

Fafchamps, Marcel. 2003. Risk, Poverty and Development. Elgar Publishing.

Ford Foundation. 2000. Roundtable on Microinsurance Services in the Informal Economy: The Role of
Microfinance Institutions.

Gaviria, Alejandro and Carmen Pages. 1999. Patterns of Crime Victimization in Latin America. Research
Department Working paper 408. Inter-American Development Bank.

Genicot and Ray. 2003. Group Formation in Risk-Sharing Arrangements. Review of Economic Studies, 70
(1), pp. 87-113.

Ibarra, Hector. 2003. Comments on ‘Risk Managed Challenges in Rural Financial Markets’ by Jerry
Skees. Mimeo. AGROASEMEX.

Jacoby, Hanan G. 1994. Borrowing Constraints and Progress through School: Evidence from Peru. The
Review of Economics and Statistics, 76(1), pp. 151-160.

Jannson, Tor and Mark Wenner. 1997. Financial Regulation and its significance for Microfinance in Latin
America and the Caribbean. Sustainable Development Department working paper. Inter-
American Development Bank. Washington D.C.

Holzmann, Robert and Steen Jørgensen. 2000. .Social Risk Management: A New Conceptual Framework
for Social Protection and Beyond. Social Protection Discussion Paper Series no. 6. Washington
D.C.: The World Bank.

Kaztman, Ruben. 2003. La dimensión espacial en las políticas de superación de la pobreza urbana. Serie
Medio Ambiente y Desarrollo, 59. ECLAC.

Lustig, Nora (ed.). 2000. Shielding the Poor. Social Protection in the Developing World. Brookings Insti-
tution and the Inter-American Development Bank.

McCord, Michael J., Jennifer Isern and Syed Hashemi. 2001. Microinsurance: A Case Study Example of
the Full Service Model of Microinsurance Provision. Self-Employed Women’s Association. Mi-

Morduch, Jonathan. 1995. Income Smoothing and Consumption Smoothing. Journal of Economic
Perspectives, Vol. 9 (Summer), pp. 103–14.

Morduch, Jonathan. 1999. Between the State and the Market: Can Informal Insurance Patch the Safety
Net? World Bank Research Observer, Vol. 14, No.2 (August), pp. 187–207.

Morduch, Jonathan. 2004. Consumption Smoothing Across Space: Testing Theories of Risk-Sharing in
the ICRISAT Study Region of South India. In Insurance against Poverty, ed. S. Dercon. Oxford
University Press.

Moser, Caroline. 1998. The Asset Vulnerability Framework: Reassessing Urban Poverty Reduction
Strategies. World Development, 26(1), pp. 1-19.

Pizarro, Roberto. 2001. La vulnerabilidad social y sus desafíos: una mirada desde América Latina. Serie
Estudios Estadísticos y Prospectivos, 6. ECLAC.

Ravallion, Martin. 2001. On the Urbanization of Poverty. Working Paper 2586. The World Bank.

Ron, Aviva. 1999. NGOs in community Health Insurance Schemes: Examples from Guatemala and the
Philippines. Social Science and Medicine 48, pp. 939-950.

Rosenzweig, Mark and Hans Binswanger. 1993. Wealth, Weather Risk and the Composition and Profit-
ability of Agricultural Investments. Economic Journal, 103, pp. 56-78.

Rosenzweig, Mark and Kenneth Wolpin. 1993. Credit Market Constraints, Consumption Smoothing, and
the Accumulation of Durable Production Assets in Low-Income Countries: Investments in Bul-
locks in India. Journal of Political Economy, 1001(2), pp. 223-244.

Rutherford, Stuart. 1999. The Poor and their Money. Institute for Development Policy and Management,
University of Manchester.

Sadoulet, Loic. 2002. Incorporating Insurance Provisions in Microfinance Contracts. Learning from
Visa®? Discussion Paper 2002/56. WIDER.

Sen, Amartya. 1966. Peasants and Dualism with or without Surplus Labour. Journal of Political Econ-
omy, 74(5), pp. 425-450.

Skees, Jerry, Panos Varangis, Donald Larson and Paul Siegel. 2004. Can Financial Markets be Tapped to
Help Poor People Cope with Weather Risks? In Insurance against Poverty, ed. S. Dercon. Oxford
University Press.

Tesliuc, Emil D., and Kathy Lindert. 2002. Vulnerability: A Quantitative and Qualitative Assessment.
Guatemala Poverty Assessment Program. The World Bank.

Townsend, Robert M. 1995. Consumption Insurance: An Evaluation of Risk-Bearing Systems in Low-
Income Economies. Journal of Economic Perspectives, Vol. 9 (Summer), 83–102.

World Bank. 1999. Consultations with the Poor. National Synthesis Reports (Argentina, Bolivia, Brazil,
Ecuador, Jamaica). Informal publications series prepared for Global Synthesis Workshop: Con-
sultations with the Poor, Poverty Group, PREM, Washington, D.C.

World Bank. 2000. World Development Report 2000/01: Attacking Poverty. Washington, D.C.: The
World Bank.

World Bank. 2003. Guapa: The Guatemala Poverty Assessment. Washington, D.C.: The World Bank.

Zaffaroni, Cecilia. 1999. Los recursos de las familias urbanas de bajos ingresos para enfrentar situacio-
nes críticas. In Activos y estructuras de oportunidades. Estudios sobre las raíces de la vulnerabi-
lidad social en Uruguay, ed. R. Kaztman. ECLAC and UNDP.